The Impact of Oil Boom and Bust Cycles on Western North Dakota

SURLC 18-007 The Impact of Oil Boom and Bust Cycles on Western North Dakota

Prepared by: Del Peterson Ali Rahim Taleqani

Small Urban and Rural Transit Center Upper Great Plains Transportation Institute North Dakota State University Fargo, ND

October 2018

Acknowledgements Funds for this study were provided by the Small Urban and Rural Livability Center (SURLC), a partnership between the Western Transportation Institute at Montana State University and the Upper Great Plains Transportation Institute at North Dakota State University. The Center is funded through the U.S. Department of Transportation’s Office of the Assistant Secretary of Research and Technology as a University Transportation Center. The Small Urban and Rural Transit Center within the Upper Great Plains Transportation Institute at North Dakota State University conducted the research. Cover Photo Credit: “Williston, North Dakota 10-18-2008” by Andrew Filer from Seattle (ex-Minneapolis).

Disclaimer

The content presented in this report is the sole responsibility of the Small Urban and Rural Transit Center, the Upper Great Plains Transportation Institute, and the authors.

NDSU does not discriminate in its programs and activities on the basis of age, color, gender expression/identity, genetic information, marital status, national origin, participation in lawful off-campus activity, physical or mental disability, pregnancy, public assistance status, race, religion, sex, sexual orientation, spousal relationship to current employee, or veteran status, as applicable. Direct inquiries to Vice Provost for Title IX/ADA Coordinator, Old Main 201, NDSU Main Campus, 701-231-7708, [email protected].

ABSTRACT

Discoveries of shale gas reserves along with the development of horizontal drilling and hydraulic fracturing techniques, as well as the initiative to move the United States toward greater energy independence led to the most recent oil boom in western North Dakota (U.S. Congress 2007). Overall, the boom brought a billion-dollar surplus for the state budget in North Dakota from severance taxes and extensive mineral rights. The population doubled in some areas, e.g., the city of Williston grew by 67% from 2010 to 2014 (Scheyder 2016). After almost a decade of production of North Dakota shale oil, prices dropped from about $100 a barrel in 2014 to about $30 in early 2016 and the industry went into the bust portion of the economic cycle where oil production and affiliated employment and spending contracted and where the industry and state are waiting for crude oil prices to rise again (Scheyder 2016).

Transit livability index measures showed large increases from 2008 to 2012 followed by overall corrections from 2013 to 2016. These measures declined as oil production and economic advances diminished. Further, transit fleet size failed to increase with population growth, and pedestrian safety has become a concern along both rural and city highways. System dynamics simulations focused on potential mode shifts from private automobile to transit, finding that seemingly small shifts (1-2%) from auto to transit would result in millions of dollars of fuel savings in the oil patch alone.

Various models of either fixed-route or flex-route busing should be considered by transit agencies and local policy makers for the larger communities of Williston and Dickinson, while more rural providers need to update their fleets to meet demand as well. Because of recent cutbacks in state and local funding, agencies should also strive to better coordinate services while continuing to provide rides to county population centers that offer goods and services required by rural residents.

TABLE OF CONTENTS

1. INTRODUCTION ...... 1 1.1 Objectives...... 1 1.2 Organization of Content...... 1 2. LITERATURE REVIEW ...... 2 2.1 Boom ...... 2 2.2 Bust ...... 3 2.3 Rebound ...... 4 2.4 Livability ...... 5 2.5 System Dynamics ...... 6 3. RESEARCH DATA AND METHODOLOGY ...... 8 3.1 Transit Livability Data and Methodology ...... 8 3.2 System Dynamics Data and Methodology ...... 9 3.2.1 Model ...... 9 3.2.2 Population sub-model ...... 10 3.2.3 Transportation demand sub-model ...... 13 3.2.4 Economy sub-model ...... 13 3.2.5 Transportation supply sub-model ...... 14 4. INDEX AND SIMULATION RESULTS...... 16 4.1 Livability Index Results ...... 16 4.2 System Dynamics Simulation Results ...... 21 4.2.1 Scenario 1 ...... 21 4.2.2 Scenario 2 ...... 22 4.2 Summary ...... 22 5. SUMMARY AND CONCLUSIONS ...... 23 Bibliography ...... 24

LIST OF TABLES

Table 3.1 Livability Principles ...... 9 Table 3.2 Livability’s Relationship to Transit and Measurements ...... 9 Table 3.3 Population Trends by City ...... 10 Table 4.1 County Population ...... 17

LIST OF FIGURES

Figure 1.1 Nine County Study Region ...... 1 Figure 2.1 Bakken Formation ...... 2 Figure 2.2 Vertical vs. Horizontal Drilling (Curtis 2011) ...... 3 Figure 2.3 Historical Oil Prices...... 4 Figure 2.4 ND Oil Production ...... 5 Figure 2.5 Livability Qualities of Life ...... 5 Figure 2.6 Supply-demand model ...... 7 Figure 3.1 Nine-county region ...... 8 Figure 3.2 Population Estimates ...... 11 Figure 3.3 Cause-and-effect loop ...... 12 Figure 4.1 Western North Dakota Transit Ridership ...... 16 Figure 4.2 Transit Livability Indexes, 2008-2016 ...... 17 Figure 4.3 Livability Indexes for Williams and Stark counties ...... 18 Figure 4.4 Livability Indexes for Mountrail, McKenzie, and Dunn counties ...... 18 Figure 4.5 Livability Indexes for Divide, Burke, Golden Valley, and Billings counties ...... 19 Figure 4.6 Support Existing Communities ...... 19 Figure 4.7 Enhance Economic Competitiveness ...... 20 Figure 4.8 Value Communities and Neighborhoods ...... 20 Figure 4.9 Trips saved by mode shift ...... 21 Figure 4.10 Supply-demand ratio by change in investment ...... 22

1. INTRODUCTION

The western half of North Dakota has experienced tremendous change in recent years due to oil exploration and drilling. Transportation issues including congestion and road quality initially affected the area during the expansion boom years, but after almost a decade of producing North Dakota shale oil, prices dropped from around $100 a barrel in 2014 to about $30 in early 2016 and the industry went into the bust portion of the economic cycle where oil production and affiliated employment and spending contracted. This has been followed by state and local funding cutbacks that have led to uncertainty regarding economic and social conditions. Local businesses have experienced tremendous fluctuations in sales leading to uncertainty and delayed growth plans as well.

1.1 Objectives This study was conducted in western North Dakota (Figure 1.1) to analyze topics focusing on public transportation and vehicle mode choice. This research builds on previous work done by Peterson and Ndembe (2015). Objectives included determining the impact of oil boom and bust cycles on transit ridership for individuals living in the oil patch as well as how variables such as income, land use, population, and local operating investments affect livability. System dynamic models were also applied to analyze mode share and potential transportation investments in western North Dakota.

Figure 1.1 Nine County Study Region 1.2 Organization of Content

The study begins with a literature review including research focusing on oil boom and bust cycles as well as livability and system dynamic overviews. Following the literature review is an overview of the research methodology used in this study. Next, results from both livability index calculations and system dynamics simulations are discussed. Finally, an overall summary provides recommendations based on the research findings.

2. LITERATURE REVIEW

2.1 Boom

Oil booms, periods of rapid growth in oil production and affiliated employment and economic activity, are familiar to the U.S. oil industry. The earliest oil boom was the Pennsylvanian oil rush in 1859 during which the region produced a third of the world’s oil at the peak of the boom (E.P. Corporation 2018). Soon after the Ohio oil rush in the 1880s and 1890s, Texas experienced an oil boom in the early 1900s. Oil booms are not limited to the United States They have also occurred in Mexico, Canada and other regions. Discoveries of shale gas reserves along with the development of horizontal drilling and hydraulic fracturing techniques, as well as the initiative to move the United States toward greater energy independence led to the most recent oil boom in western North Dakota (Figure 2.1) (U.S. Congress 2007).

Figure 2.1 Bakken Formation

Figure 2.2 illustrates the hydraulic fracking technique compared to the traditional vertical drilling technique. Vertical drilling uses relatively shallow depths to extract oil while hydraulic fracking utilizes greater depth and horizontal drilling to reach oil and gas deposits. Because of the tremendous depth of the drill, often more than 10,000 feet in the Bakken Formation, it must also include a vertical well, thus resembling the letter “L” in the figure.

Figure 2.2 Vertical vs. Horizontal Drilling (Curtis 2011)

Overall, the boom brought a billion-dollar surplus to the state budget in North Dakota from severance taxes and extensive mineral rights. The population doubled in some areas, e.g., the city of Williston grew by 67% from 2010 to 2014 (Scheyder 2016).

Demographics changed because newcomers were mainly single men without families, and quality of life decreased as resources were exhausted in response to this transition. Pedestrian safety, for example, became a concern for those who walk for exercise along rural roads without paved shoulders. Inadequate walkways are a major problem related to pedestrians involved in motor vehicle crashes (NHTSA 2018). On the other hand, the boom resulted in a massive investment in the infrastructure in western North Dakota. Williston made a significant investment in infrastructure and amenities to catch up with the growth and prepare for the next boom. Expenditures included: $110 million for water- treatment plants, $56 million for a high school, $20 million for a jail expansion, and $70 million for a recreational center. North Dakota’s Legacy Fund is now worth $3.5 billion, prompting debate over spending and management of the fund. These figures are based on records officially available as of June 30, 2017 (Scheyder 2016).

2.2 Bust After almost a decade of production of North Dakota shale oil, prices dropped from around $100 a barrel in 2014 to about $30 in early 2016, and the industry went into the bust portion of the economic cycle where oil production and affiliated employment and spending contracted and is the industry and state are waiting for crude oil prices to reach $60 per barrel again (Figure 2.3). There is, to some extent, an analogy between the Klondike gold rush and the western North Dakota oil boom. From an economic perspective, there have been two competing dynamics: 1) Positive spillover as the oil industry fuels labor demand and a broad variety of other economic activity; and 2) Crowding-out effect which refers to a disruption in the labor market. The oil industry offers good-paying jobs and other industries have no alternative but to raise wages. The Williston Walmart increased the base pay to $17 per hour, which was among the highest pay rates of any retailer in the country. This hurt companies in other sectors such as manufacturing and agriculture as they had to compete in markets with aggressive competitors (Wegmann 2014).

120

100

60 US $

0 1/2/2009 1/2/2010 1/2/2011 1/2/2012 1/2/2013 1/2/2014 1/2/2015 1/2/2016 1/2/2017 Date

Figure 2.3 Historical Oil Prices 2.3 Rebound

The boom crashed as a result of a boost in oil production in countries like Saudi Arabia and Russia, causing significant financial losses and a contraction in the number of jobs. The situation, however, has recently begun to change. There are some signs of an unstable rebound in the oil industry. As of March 2018, there were 51 drilling rigs in North Dakota, up from a low of 27 in May 2016, but somewhat below the high of 218 in December 2012. Monthly oil production oscillated around the 1 million barrels-a-day mark during 2017, down from a peak of 1.23 million in December 2014 (Figure 2.4). Production is expected to be supported by completion of the Dakota Access pipeline, which is reducing the cost of oil transportation from the Bakken. In November 2017, OPEC also announced plans to reduce production, which tends to boost oil prices. However, oil inventories are still historically high. A price of $60 per barrel of oil is an attractive investment in the Bakken area. Finally, recent employment reports have showed larger employment in the Bakken area than previously thought. The local industry has also become more efficient, and new methods to forecast employment may be needed (Bangsund and Hodur 2017).

1400

1200

1000

800 Daily 600 Output in

Barrels in 000s Barrels 400

200

0 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Year

Figure 2.4 ND Oil Production 2.4 Livability

Livability is defined as suitability for human living according to the Merriam-Webster dictionary (Merriam-Webster 2018). Additionally, according to a TCRP research report, livability is “people having good access to opportunities they can use in the pursuit of improvements to their quality of life” (Transportation Research Board 2016). It is a metric for measuring quality of life (QoL) of a given geographical place. Livability is the subject of many studies that address objective conditions in a quantitative way (Veenhoven and Ehrhardt 1995, Lyubomirsky et al. 2005, Anderson and Van Kempen 2003). Such a ranking is just a number used to standardize the favorability scale of a place objectively. However, subjectivity is missing those rankings. In other words, one might live in a city with the highest ranking, but have no positive feelings toward that city’s quality of life. Kozaryn described the difference between rankings and perceptions by categorizing three levels of quality of life; 1) normative 2) subjective 3) objective (Figure 2.5) (Okulicz-Kozaryn 2013). He found that the relationship between subjective and objective measures is weak and concluded that perception is much more important than the ranking system.

Normative

Objective Subjective

Figure 2.5 Livability Qualities of Life

2.5 System Dynamics

System dynamics (SD) is a powerful tool for analyzing large-scale, complex socio-economic systems. This methodology has been applied to various fields, including, but not limited to, global environmental sustainability, regional sustainable development issues, environmental management, water resource planning and ecological modeling, agricultural sustainability, regional environmental planning and management, national development programs, transportation, and land use.

All SD models have three types of variables: stock, rate, and auxiliary; and two kinds of flows: physical/material and information. Variables could interact with each other only through those two flows. Variables, together with flows, provide the fundamental structure of one dynamic system, called the stock-flow diagram. In SD, simulation is governed entirely by the passage of time known as ‘‘time- step’’ simulation. The common purpose of this approach is to understand the fundamentals of the dynamics of concern and to search for managerial decisions to improve the situation. These dynamics refer to the long-term, macro-level decision rules used by upper management. Figure 2.6 illustrates the general concept of the SD model incorporating both transportation supply and demand.

Abbas and Bell (1994) described the advantages of using a SD approach to transportation modeling as follows: 1. A systematic approach to the transportation problem 2. Feedback interaction between supply and demand vs. supply/demand equilibrium 3. Integrated and comprehensive view incorporating all other related subsectors vs. conventional 4. SD models capture the dynamic of the transportation system and generate results dynamically 5. SD models produce experimental tools, providing more flexibility for future analysis However, there are limitations of SD 1. Reaching an optimal design is difficult. Some research has developed a heuristic optimization approach, but it is tedious work. 2. It is also difficult to validate the results of the SD model because transportation planners evaluate the usefulness of the results rather than numerical results.

Travel by Transit Person Trip Transportation Demand Travel by Private Vehicle Supply-Demand Ratio

Road Capacity Transportation Supply

Figure 2.6 Supply-demand model

Rodrigue et al. (2017) describe the transportation system as a collection of several elements such as infrastructures, modes, and terminals, enabling individuals, institutions, corporations, regions, and nations among others. Such a system supports and drives the mobility of people, freight, and information. Mobility must occur over infrastructures with a fixed capacity which translates to transportation supply. In other words, the transportation system consists of two main components: travel demand and transportation supply (Rodrigue et al. 2017). On the one hand, transportation supply is the capacity of transportation infrastructures and modes over a geographic area for a specific period. Practitioners usually express the capacity in terms of infrastructures (capacity), services (frequency) and networks (coverage). On the other hand, transport demand is the need for mobility. Similar to transport supply, it can be expressed as number of people, volume, or tons per unit of time and space.

3. RESEARCH DATA AND METHODOLOGY

Data and methodology for both the livability index and system dynamics models are the focus of this chapter. Calculations were conducted for the nine-county region of North Dakota most heavily impacted by oil production fluctuations (Figure 3.1). The nine counties included in the index were Divide, Burke, Williams, Mountrail, McKenzie, Golden Valley, Billings, Dunn, and Stark. Both the livability index and system dynamics models included historical data calculations as well as forecasted scenarios dependent on regional estimates.

Figure 3.1 Nine-county region

3.1 Transit Livability Data and Methodology Transit livability indexes were calculated based on six core livability principles developed by the Partnership for Sustainable Communities (Table 3.1) (DOT, HUD, and EPA 2014). Data used to calculate index measures were collected from both the National Transit Database (Federal Transit Administration 2016), and the American Community Survey (U.S. Census 2016).

Table 3.1 Livability Principles Provide more transportation choices Promote equitable, affordable, housing Enhance economic competitiveness Support existing communities Coordinate policies and leverage investment Value communities & neighborhoods

Table 3.2 lists each livability principle and its relationship to public transportation. Previous work by both Peterson and Ndembe (2015) and Brooks et. al. (2013) focused on the relationship between transit and livability. These relationships are defined in this research as well as in those previously mentioned. The transit livability index measures used in this study are shown in the third column.

Table 3.2 Livability’s Relationship to Transit and Measurements Livability Principle Relationship to Transit Index Measure Transit service provides an alternative Percent of workers that do Provide more transportation choices transportation choice. not drive alone to work Household income after Promote equitable, affordable Transit provides a means to connect home transportation and housing housing owners to communities and can lower overall housing and transportation expenses. expenses Transit provides greater accessibility to workers for commuting and access to services, Revenue vehicles/county Enhance economic competitiveness improving the economic competiveness of a population community. Transit utilizes the existing built environment Ridership/developed land Support existing communities to serve and support an existing community. area State and local operating Coordinate policies and leverage Transit coordinates funding from federal, investment/operating investment state, and local entities to provide quality service and operate cost-effectively. expenses Transit adds value to local communities by Ridership/county mobility Value communities & neighborhoods serving local residents who deserve safe, affordable transportation choices while often needs index possessing mobility disadvantages.

All transit livability index measures were calculated at the county level from 2008 to 2016. Time series data illustrates the impact of oil production on transit during the defined time period. Equal weighting was assigned to all measures as individual index measures illustrate a specific livability principle. Thus, no specific measure was quantified as being more important than any other. Also, after initial calculations were completed, the index was categorized in percentiles from 1 to 10 using a normal distribution. This provided consistency for analysis and comparison among results. 3.2 System Dynamics Data and Methodology

3.2.1 Model In this section, we identified the variables and established the relevant equations based on the feedback and cause-and-effect loops as shown in Figure 3.3. We limited the scope of this part of the study to the

9 cities located in nine North Dakota counties to address two dominate modes of transportation (personal vehicle vs. public bus). These cities are home to 62% of the entire population in the counties (U.S. Census Bureau 2012). The rest of population lives in very small towns or rural areas with little to no access to transit service.

3.2.2 Population sub-model

The population sub-model reflects the developing stage of selected cities in this model (Table 3.3). Population size influences the total transportation demand. We assumed the average population growth rate of 5% inclusive of birth rate, death rate, and net migration rate, which is reasonably close to official estimates (Figure 3.2). • Population = populationGrowthRate * Population

Table 3.3 Population Trends by City City County Population 2000 Population 2010 Population 2017 Alamo Williams 51 57 69 Alexander McKenzie 217 223 308 Ambrose Divide 23 26 27 Arnegard McKenzie 105 115 152 Beach city Golden Valley 1,116 1,019 1,065 Belfield Stark 866 800 976 Bowbells Burke 406 336 340 Columbus Burke 151 133 139 Crosby Divide 1,089 1,070 1,310 Dickinson Stark 16,010 17,787 22,186 Dodge Dunn 125 87 100 Dunn Center Dunn 122 146 179 Epping Williams 79 100 115 Flaxton Burke 73 66 66 Fortuna Divide 31 22 22 Gladstone Stark 248 239 351 Golva Golden Valley 106 61 68 Grenora Williams 202 244 287 Halliday Dunn 227 188 197 Killdeer Dunn 713 751 1,144 Lignite Burke 174 155 230 Medora Billings 100 112 132 New Town Mountrail 1,367 1,925 2,528 Noonan Divide 154 121 119 Palermo Mountrail 77 74 85 Parshall Mountrail 981 903 1,250 Plaza Mountrail 167 171 198 Portal Burke 131 126 143

Powers Lake Burke 309 280 284 Ray Williams 534 592 780 Richardton Stark 619 529 538 Ross Mountrail 48 97 113 Sentinel Butte Golden Valley 62 56 63 South Heart Stark 307 301 410 Springbrook Williams 26 27 32 Stanley Mountrail 1,279 1,458 2,645 Taylor Stark 150 148 163 Tioga Williams 1,125 1,230 1,499 Watford City McKenzie 1,435 1,744 6,523 White Earth Mountrail 63 80 91 Wildrose Williams 129 110 131 Williston Williams 12,512 14,716 25,586 Total 43,709 48,425 72,644

180000

160000

140000

120000

100000

80000

60000

40000

20000

0 1995 2000 2005 2010 2015 2020 2025 2030

Figure 3.2 Population Estimates

Figure 3.3 Cause-and-effect loop

3.2.3 Transportation demand sub-model

Transportation demand is heavily dependent on population. Transportation demand is the sum of the demand for transit and private car and can be expressed as: 𝐷𝐷 = + _

𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑐𝑐𝑐𝑐𝑐𝑐 The variables of this sub-model include𝐷𝐷 the average𝐷𝐷 daily𝐷𝐷 person trips, trips by transit, trips by private car, the demand for transit, the demand by private car and the total demand for transportation. We used data from 2009 National Household Travel Survey for person trips per day, mode share by transit, and mode share by private car (U.S. DOT 2011) . According to the 2017 Rural Transit Fact Book, the average seating capacity for buses and automobile are 26.4 and 4.32 persons respectively. Given level terrain for city streets, passenger car equivalent (PCE) for buses varies with flow rates of 1.7, 1.2, 1.1 for less than 600, between 600 and 1200, more than 1200 passenger car per hour respectively, according to the Highway Capacity Manual (HCM 2010). The average daily number of trips is 3.79 trips per person1. The equations of this sub-model are:

Population growth rate POR Person trips per year PTY Person trip per day PTD2 Population (annual) POP Trips by transit TBT Trips by private car TBC Mode share by transit MST Mode Share By Private Car MSC Average seating capacity by transit SCT Average seating capacity by private car SCC Passenger car equivalent for transit PCE

• POPannual = POR * POP • PTY = POPannual * PTD * 365 • TBT = PTY * MST • TBC = PTY * MSC • DBT = (TBT / SCT) * PCE • DBC = TBC / SCC • TRD = DBT + DBC 3.2.4 Economy sub-model The economy sub-model reflects the forces driving rural transportation development. The level of the economic development is one of the city competitiveness indicators and directly affects transportation investment because the government has to increase spending on transportation infrastructure to

1 https://nhts.ornl.gov/2009/pub/stt.pdf - Table 3 2 National Household Survey defined person trip as a trip from one address to another by one or more persons in any mode of transportation. Each person is considered as making one person trip. For example, four persons traveling together in one auto are counted as four person trips https://nhts.ornl.gov/2009/pub/stt.pdf.

13 accommodate traffic. We defined GDP as a level variable (annual), annual GDP growth rate as a rate variable, and transportation investment rate as an auxiliary variable. Because GDP was not available at the county level, we used the GDP values from Bureau of Economic Analysis (2018) for calculating GDP growth rate. For the base year, we used the per capita GDP of North Dakota multiplied by the base year population (U.S. Bureau of Economic Analysis 2018). For transportation investment rate, we considered the investment data as percent of GDP from the Congressional Budget Office report (Musick and Petz 2015). The annual transportation investment rate follows a uniform distribution with a mean of 1.03 percent. The equations of this sub-model are as follows:

Gross domestic product GDP GDP per capita growth rate GCR Transportation investment as percentage of GDP TIP Transportation investment TRI

• GDPannual = GCR * GDP • TRI = GDP * TIP 3.2.5 Transportation supply sub-model Transportation supply reflects the level of built rural infrastructure and maintains a dynamic equilibrium with transportation demand. The level of transportation supply depends on the investment in infrastructure construction and improvement. The equations of this sub-model are as follows:

Lane miles LMI Cost per lane mile CLM Lane mile growth rate LMR Passengers cars PAC Traffic flow TRF Passenger car equivalent per hour PCH Transportation supply TRS Supply to demand ratio SDR

• LMI = TRI / CLM • PAC = PCH * 24 * 365

The North Dakota Transportation Handbook estimates that it costs almost $2 million per lane-mile for total reconstruction including grading and asphalt surfacing. However, this cost does not cover all construction-related costs. To estimate traffic flow rate (in passenger cars per hour per lane) on a daily basis, we use the 2010 Highway Capacity Manual speed-flow curve for a speed limit of 25 mph at different levels of service (H.C. Manual 2010). The transportation supply is the capacity of city streets in terms of the number of passenger car equivalent. 𝑆𝑆

Finally, the transportation supply-demand ratio used to quantitatively describe the transportation system state is the ratio between transportation supply and travel demand. As the value of this ratio increases, rural transportation would become less obstructed. Figure 2.6 shows the supply-demand ratio estimating framework. We define as the transportation supply-demand ratio where and are transport supply and travel demand, the model can be written as: 𝛼𝛼 𝑆𝑆 𝐷𝐷 = 𝑆𝑆 𝛼𝛼 �𝐷𝐷

4. INDEX AND SIMULATION RESULTS 4.1 Livability Index Results

Transit agencies in the North Dakota oil boom region saw tremendous ridership growth from 2008 to 2011. Ridership remained at or above 100,000 rides per year through 2014, but recently fell to 80,000 rides per year in 2016 (Figure 4.1). The largest decreases in ridership from 2014 to 2016 occurred in Williams and Stark counties. These are the two most populated counties in the nine-county oil boom region and their economies have been greatly affected by the recent decline in oil prices. However, comparing 2008 to 2016 ridership shows that overall ridership still nearly doubled during this time frame.

120,000 110,000 100,000 90,000 80,000

Ridership 70,000 60,000 50,000 40,000 2008 2009 2010 2011 2012 2013 2014 2015 2016 Year Figure 4.1 Western North Dakota Transit Ridership

Figure 4.2 shows the average transit livability index results from 2008 to 2016. The county indexes were classified ranging from 1 to 10, with Mountrail County showing the highest index value for the time period while Divide County generated the lowest. Overall, index values ranged from 4.34 to 5.70 which was relatively consistent considering the differing community sizes, demographics, and locations of the counties relative to the center of the oil patch. All raw data calculations can be found in the Appendix.

Billings 5.38 Golden Valley 5.11 Burke 5.04 Divide 4.34 Dunn 4.83 McKenzie 4.83 Mountrail 5.70 Stark 5.40 Williams 5.06

1 2 2 3 3 4 4 5 5 6 6 Figure 4.2 Transit Livability Indexes, by County, 2008-2016

The time series livability index results were categorized by county population. Table 4.1 shows the nine counties studied with their respective populations. The third column shows the categorization made for the analysis. Comparing counties with populations greater than 30,000 to those with fewer than 2,000 was not practical. For example, a relatively small change in either transit ridership or county population had much more influence on index values compared to a similar change among counties with larger populations. For this main purpose, and also because illustrating nine counties within one figure was both confusing and cumbersome, counties were segregated. Table 4.1 County Population County Population Size Williams County 33,349 Large counties Stark County 30,209 McKenzie County 12,724 Medium counties Mountrail County 10,265 Dunn County 4,289 Divide County 2,288 Small counties Burke County 2,131 Golden Valley County 1,789 Billings County 940

Figure 4.3 illustrates the nine-year transit livability indexes for Williams and Stark counties. Both counties showed an overall increase in their respective indexes from 2008 to 2013 primarily because of rapid population and transit ridership increases. However, from 2014 to 2016 their respective indexes have leveled off and have even begun to decrease. This is, undoubtedly, because of to the downturn in oil production. For example, in Williams County alone, transit ridership fell by roughly 40% between 2013 and 2016. Average income levels, however, have increased substantially since 2008, resulting in only recent minor declines within the overall index.

6.0

5.5

5.0 Williams

Index Value Index Stark 4.5

4.0 2008 2009 2010 2011 2012 2013 2014 2015 2016 Year Figure 4.3 Transit Livability Indexes for Williams and Stark counties

Figure 4.4 shows the transit livability indexes for Mountrail, McKenzie, and Dunn counties. A combination of increased funding from state and local sources along with population growth, and a substantial increase in household income caused a significant increase in index values from 2008 to 2013. However, primarily because of the downturn in the oil patch, local funding and household incomes have decreased and population growth has stagnated from 2014 to 2016. For example, Dunn County has seen a decrease of nearly 15% in real household income in just 2 years, between 2014 and 2016, while Mountrail County has realized similar decreases as well.

6.5

6.0

5.5 Mountrail 5.0 McKenzie Index Value Index 4.5 Dunn 4.0

3.5 2008 2009 2010 2011 2012 2013 2014 2015 2016 Year Figure 4.4 Transit Livability Indexes for Mountrail, McKenzie, and Dunn counties

Figure 4.5 shows transit livability indexes for Divide, Burke, Golden Valley, and Billings counties. These results are not as easy to interpret as those from larger counties. All of the indexes peaked for each county represented from 2013 to 2015. Overall transit ridership declined in Divide County while remaining relatively constant in the other three counties during the past few years. Also, real household incomes, along with state and local funding, have not fluctuated substantially compared to the larger counties in the oil patch. Overall, the smaller counties have not seen the dramatic boom and bust cycles from oil production that have been present in larger counties, resulting in lower volatility among index values.

6.0

5.5

5.0 Divide 4.5 Burke

Index Value Index 4.0 Golden Valley Billings 3.5

3.0 2008 2009 2010 2011 2012 2013 2014 2015 2016 Year Figure 4.5 Transit Livability Indexes for Divide, Burke, Golden Valley, and Billings counties

The previous figures highlight livability indexes averaged across all six livability measures. Disaggregating the data to look more closely at individual measures yields some interesting findings. The individual livability measures not mentioned below can be found in Appendix A. Figure 4.6, for example, shows the index values for supporting existing communities. This measure is calculated by dividing transit ridership by the developed land area of a given county. Notice that values increased dramatically from 2008 to 2011 as ridership increased while the developed land area remained relatively constant. However, dramatic land development, especially within the larger counties of Stark and Williams coupled with decreased ridership, caused this measure to drop significantly from 2011 to 2016. This also raises concerns for pedestrian safety as more land is developed around communities and traffic begins to increase. Coupled with oil production traffic, pedestrians now have fewer options when contemplating walking for exercise along once primarily rural roads and must choose either different exercise alternatives, or relatively unsafe walking conditions.

7.0 6.5 6.0 5.5 5.0 Large Counties 4.5 Medium Counties 4.0 3.5 Small Counties 3.0 2.5 2.0 2008 2009 2010 2011 2012 2013 2014 2015 2016

Figure 4.6 Support Existing Communities

Results from the livability measure for enhancing economic competitiveness also raise concern. Figure 4.7 illustrates that for, both small and large counties, this index measure has fallen from its 2011 high. It is calculated by dividing the number of transit vehicles by the population they serve. Results indicate that as populations increased in recent years throughout the oil producing region, the number of transit vehicles has not increased proportionally. This is predominantly true among larger counties which have seen tremendous gains in total population with little to no increase in transit vehicles to serve their populations.

7.0 6.5 6.0 5.5 Large Counties 5.0 4.5 Medium Counties Index Value Index 4.0 Small Counties 3.5 3.0 2008 2009 2010 2011 2012 2013 2014 2015 2016 Year

Figure 4.7 Enhance Economic Competitiveness

Figure 4.8 shows the index measure for valuing communities and neighborhoods. This variable is calculated by dividing transit ridership by the mobility needs index of a given county. Once again, this particular measure has steadily decreased or remained constant for all counties during the past three to four years. This corresponds to decreased ridership seen in Figure 4.1 as well as an increasing mobility needs index among both large- and medium-sized counties. Mobility needs indexes have grown along with the increasing populations of these counties while the largest counties are showing needs indexes approaching those in some of North Dakota’s most heavily populated areas.

9.0

8.0

7.0

6.0 Large Counties 5.0 Medium Counties Index Value Index 4.0 Small Counties

3.0

2.0 2008 2009 2010 2011 2012 2013 2014 2015 2016 Year Figure 4.8 Value Communities and Neighborhoods

4.2 System Dynamics Simulation Results

In this section, two different scenarios are considered including: 1) change in mode share; 2) increase in transportation investment as a percentage of GDP by checking the sensitivity of the supply-demand ratio.

4.2.1 Scenario 1

As expected, because of several generalizations throughout the model development and the nature of rural transportation system, there is excess supply throughout the forecasting period from 2017 to 2027. However, the percentage change was checked by shifting the mode share from private cars to transit by 0.5, 1, 1.5, and 2%. The percentage change corresponding to the supply-demand ratio change are 0.497%, 0.998%, 1.505%, and 2.016% for 0.5%, 1%, 1.5%, and 2% change in mode shift. The change in supply-demand ratio reasonably corresponds to percentage change for the mode shift from private cars to transit alternative.

On the other hand, as shown in Figure 4.9, the number of trips saved by such a mode shift increased relatively linearly. For example, in year 2018, almost 220, 440, 660, 880 thousand trips might have been saved by 0.5%, 1%, 1.5% 2% mode shifts, respectively. Given a 36.13 mile average trip length and an average fuel efficiency of 23.4 miles per gallon for private vehicle type, almost 1.36 million gallons of fuel would be saved with a mode shift of only 2%.

Number of Trips Saved by Mode Shift -

(100,000.00) 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027

(200,000.00)

(300,000.00)

(400,000.00)

(500,000.00) Trips (600,000.00)

(700,000.00)

(800,000.00)

(900,000.00)

(1,000,000.00)

0.05% 1% 1.5% 2%

Figure 4.9 Trips saved by mode shift

4.2.2 Scenario 2

In scenario 2, the investment as the percentage of GDP from the base year mean value of 1.03 was increased by 5% each year. Surprisingly, as shown in Figure 4.10, the impact of a change in mode shift is much more significant on supply-demand ratio than the change in investment. With 20% percent change increase in investment, the supply-demand ratio changes only 1.4 percent.

Supply-Demand Ratio Change Due to the Change in Investment 1.6000% 1.4000% 1.2000% 1.0000% 0.8000% 0.6000% 0.4000% 0.2000% 0.0000% 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027

5.0% 10.0% 15.0% 20.0%

Figure 4.10 Supply-demand ratio by change in investment 4.2 Summary

The main objective of the transit livability index measures in this chapter was to determine the effect of the western North Dakota oil boom on livability in the region, especially as it pertains to public transportation. Overall, measures have been decreasing in recent years as oil production and economic advances have diminished. Fewer workers are utilizing transit and more commuters are commuting alone to work on a daily basis. Also, an increasing mobility needs index, primarily in Stark and Williams counties, has led to lower measurements within the valuing communities and neighborhoods index associated with communities located in these counties. Further, the size of the transit fleets within these counties has not increased proportionally to population growth from 2008 to 2016 and this has caused the measure of economic competitiveness to decrease and/or stagnate. Finally, pedestrian safety has become a concern as more land in and around local communities has been developed, and, when coupled with traffic increases from oil production, pedestrians may no longer feel safe while walking on rural roads for exercise.

System dynamics simulation findings showed that a relatively small mode shift of 2% from private vehicle to transit could have a substantial effect on oil patch traffic flows. However, the most noticeable improvement would be financial, as this same 2% mode shift was estimated to save more than 1.36 million gallons of fuel in the oil patch during the forecast time period.

5. SUMMARY AND CONCLUSIONS

Discoveries of shale gas reserves along with the development of horizontal drilling and hydraulic fracturing techniques, as well as the initiative to move the United States toward greater energy independence led to the most recent oil boom in western North Dakota (U.S. Congress 2007). Overall, the boom resulted in a billion-dollar surplus for the state budget in North Dakota from severance taxes and extensive mineral rights. The population doubled in some areas, e.g., the city of Williston grew by 67% from 2010 to 2014 (Scheyder 2016). After almost a decade of production of North Dakota shale oil, prices dropped from around $100 a barrel in 2014 to about $30 in early 2016 and the industry went into the bust portion of the economic cycle where oil production and affiliated employment and spending contracted and the state and industry are waiting for crude oil prices to reach $60 per barrel again (Scheyder 2016).

Transit livability index measures showed large increases from 2008 to 2012 followed by overall corrections from 2013 to 2016. These measures declined as oil production and economic advances diminished. Further, transit fleet size has failed to increase with population growth and pedestrian safety has become a concern along both rural and city highways. System dynamics simulations focused on potential mode shifts from private automobile to transit, finding that seemingly small shifts (1-2%) from auto to transit would result in millions of dollars of fuel savings in the oil patch.

A major finding of this research shows that although the recent oil bust has caused considerable concern in western North Dakota, the population and transit ridership are considerably larger today than they were in 2008. For example, transit ridership nearly doubled from 2008 to 2016 (Figure 4.1), due largely to the expanding local economy. Various models of either fixed-route or flex-route busing should be considered by transit agencies and local policy makers for the larger communities of Williston and Dickinson, while more rural providers need to update their fleets to meet demand as well. Policy makers should also consider that the majority of local rural transit riders are elderly and need quality vehicles with updated suspensions to provide comfortable rides for their aging clientele. Because recent cutbacks in state and local funding, agencies should also strive to better coordinate services while continuing to provide rides to county population centers that offer the goods and services many rural residents require.

Bibliography

Abbas, K. A. and Bell, M. G. H. “System dynamics applicability to transportation modeling,” Transportation Research Part A -- Policy and Practice, vol. 28A, no. 5, Sep. 1994. Andersen, H. T. and Van Kempen, R. “New trends in urban policies in Europe: evidence from the Netherlands and Denmark,” Cities, vol. 20, no. 2, pp. 77–86, 2003. Bangsund, D. A. and Hodur N. M. “Williston Basin 2016 : Employment , Population , and Housing Forecasts,” Agribusiness & Applied Economics Report No. 769, North Dakota State University, Fargo, ND, 2017. Brooks, J. M., Edrington, S., and Catala, M. “The Nexus of Livability, Transit, and Performance Measurement,” Texax A&M Transportation Institute, Presented at the the CALACT Spring Conference, May, 2013. DOT, HUD, and EPA, “Partnership for Sustainable Communities: Five Years of Learning from Communities and Coordinating Federal Investments, Fifth Anniversary Report,” p. 21, 2014. E. P. Corporation, “Early Oil in Pennsylvania.” E.P. Corportation, 2018. [Online]. Available: http://www.enopetroleum.com/earlyoilpennsylvania.html. Federal Transit Administration, “National Transit Database,” 2016. https://www.transit.dot.gov/ntd H. C. Manual, “HCM2010,” Transportation Research Board, National Research Council. Washington, DC, 2010. Lyubomirsky, S., Sheldon K.M., and Schkade,D. “Pursuing happiness: The architecture of sustainable change.,” Review of General Psychology, vol. 9, no. 2, p. 111, 2005. Merriam-Webster, “Livability | Definition of Livability by Merriam-Webster,” 2018. Musick, N. and Petz, A. “Public Spending on Transportation and Water Infrastructure, 1956 to 2014,” no. March, p. 31, 2015. National Highway Traffic Safety Administration (NHTSA), “Traffic Safety Facts,” vol. 2018, no. March, pp. 1–10, 2015. Okulicz-Kozaryn, A. “City Life: Rankings (Livability) Versus Perceptions (Satisfaction),” Social Indicators Research, vol. 110, no. 2, pp. 433–451, Jan. 2013. Peterson, D. and Ndembe, E. “The Impact of North Dakota’s Oil Boom on Transit Livability,” SURLC 15- 002, Montana State University, Bozeman: Small Urban and Rural Livability Center, 2015. Rodrigue, J.-P., Comtois C., and Slack, B. The Geography of Transport Systems, 4th ed. New York: Routledge, 2017. Scheyder, E. “In North Dakota’s oil patch, a humbling comedown,” Reuters, 2016. [Online]. Available: https://www.reuters.com/investigates/special-report/usa-northdakota-bust/. Transportation Research Board, Livable Transit Corridors: Methods, Metrics, and Strategies, TCRP Research Report 187. Washington, DC: The National Academies Press, 2016. U.S. Bureau of Economic Analysis, “Real GDP by state (millions of chained 2009 dollars),” 2018.

U.S. Bureau of Economic Analysis, “Per capita real GDP by state (chained 2009 dollars),” 2018. US Census, “American Community Survey,” 2016. https://www.census.gov/programs-surveys/acs/ U. S. Census Bureau, “North Dakota : 2010,” June, 2012. https://www.census.gov/prod/cen2010/cph-2- 36.pdf U. S. Congress, Energy independence and security act of 2007. Washington, D.C.: US Government Printing Office, 2007. U.S. Department of Transportation, “Summary of Travel Trends: 2009 National Household Travel Survey,” Federal Highway Administration., p. 82, 2011. Veenhoven. R. and Ehrhardt, J. “The cross-national pattern of happiness: Test of predictions implied in three theories of happiness,” Social Indicators Research, vol. 34, no. 1, pp. 33–68, 1995. Wegmann, P. “Why One Walmart in North Dakota Is Paying $17.40 an Hour,” The Daily Signal, 2014. [Online]. Available: https://www.dailysignal.com/2014/06/10/drilling-innovation-forcing-walmart-northdakota-pay-17-40-hour/.

APPENDIX A: INDIVIDUAL LIVABILITY MEASURES

6.5 6.0 5.5 5.0 4.5 Large Counties 4.0 Medium Counties

Index Value Index 3.5 Small Counties 3.0 2.5 2.0 2009 2010 2011 2012 2013 2014 2015 2016 Year Coordinate Policies and Leverage Investments

8.0

7.0

6.0 Large Counties 5.0 Medium Counties

Index Value Index 4.0 Small Counties 3.0

2.0 2008 2009 2010 2011 2012 2013 2014 2015 2016 Year Percent Who do not Drive Alone to Work

9.0

8.0

7.0 Large Counties 6.0 Medium Counties 5.0 Small Counties 4.0

3.0 2008 2009 2010 2011 2012 2013 2014 2015 2016

Promote Equitable and Affordable Housing

APPENDIX B: LIVABILITY MEASURES, RAW DATA Golden Williams Stark Mountrail McKenzie Dunn Divide Burke Valley Billings C&L Fed Investment 2009 38.1% 35.2% 39.8% 38.1% 37.1% 38.1% 39.8% 42.3% 42.3% 2010 40.0% 21.3% 36.1% 40.0% 39.0% 40.0% 36.1% 41.2% 41.2% 2011 41.8% 20.2% 39.3% 41.8% 40.4% 41.8% 39.3% 40.4% 40.4% 2012 18.7% 21.4% 43.6% 18.7% 44.2% 18.7% 43.6% 67.0% 67.0% 2013 40.9% 42.8% 52.8% 40.9% 52.8% 40.9% 52.8% 63.6% 63.6% 2014 43.7% 38.9% 46.8% 43.7% 43.6% 43.7% 46.8% 62.0% 62.0% 2015 41.1% 37.0% 43.4% 41.1% 48.0% 41.1% 43.4% 61.3% 61.3% 2016 37.1% 33.1% 44.5% 37.1% 36.7% 37.1% 44.5% 63.7% 63.7%

Value Communities and Neighborhoods 2008 2838 2770 1798 1787 2502 888 1061 2348 1126 2009 4026 5399 2289 2534 2535 1259 1801 1182 567 2010 6209 8982 3520 3517 2043 2185 2308 1973 946 2011 5930 11495 3492 3111 1908 2318 1718 1589 762 2012 5775 10078 3614 3029 1997 2258 1778 2318 1111 2013 6100 8881 3112 2742 2319 2385 1722 1961 940 2014 5334 7831 3311 2398 1827 2085 1832 1993 956 2015 5066 6476 3043 2278 1680 1981 1684 1946 933 2016 3634 6375 3140 1634 1807 1421 1738 2382 1142

Enhance Economic Competitiveness 2008 0.00028 0.00051 0.00042 0.00028 0.00022 0.00028 0.00042 0.00148 0.00148 2009 0.00027 0.00050 0.00042 0.00027 0.00022 0.00027 0.00042 0.00185 0.00185 2010 0.00029 0.00053 0.00046 0.00029 0.00022 0.00029 0.00046 0.00185 0.00185 2011 0.00030 0.00052 0.00048 0.00030 0.00022 0.00030 0.00048 0.00185 0.00185 2012 0.00024 0.00048 0.00042 0.00024 0.00022 0.00024 0.00042 0.00111 0.00111

2013 0.00022 0.00043 0.00042 0.00022 0.00025 0.00022 0.00042 0.00111 0.00111 2014 0.00029 0.00046 0.00040 0.00029 0.00033 0.00029 0.00040 0.00111 0.00111 2015 0.00022 0.00037 0.00040 0.00022 0.00024 0.00022 0.00040 0.00148 0.00148 2016 0.00022 0.00035 0.00038 0.00022 0.00036 0.00022 0.00038 0.00148 0.00148

Percent who do not drive alone to work 2008 14.7% 15.6% 15.0% 14.0% 20.9% 11.1% 24.8% 12.8% 19.0% 2009 12.3% 14.0% 17.1% 13.8% 23.6% 13.9% 25.0% 12.1% 15.3% 2010 11.7% 14.0% 17.6% 15.1% 20.9% 16.3% 24.3% 13.1% 20.6% 2011 14.7% 12.7% 17.6% 15.1% 20.9% 16.3% 24.3% 13.1% 20.6% 2012 14.1% 15.6% 18.1% 16.9% 22.7% 19.0% 27.2% 17.4% 24.4% 2013 14.4% 14.9% 18.1% 16.9% 22.7% 19.0% 27.0% 17.4% 24.4% 2014 12.9% 14.3% 17.0% 17.4% 22.7% 20.0% 23.7% 18.3% 29.3% 2015 15.2% 13.3% 16.0% 16.2% 20.3% 17.0% 20.2% 19.8% 28.6% 2016 15.5% 14.9% 13.6% 16.1% 19.2% 20.0% 17.6% 19.6% 24.5%

Promote Equitable Affordable Housing 2008 40,626 33,242 36,854 35,731 36,481 35,878 39,396 18,484 44,030 2009 42,897 35,196 42,755 37,402 38,844 34,405 38,982 22,092 40,056 2010 48,666 39,077 44,305 41,914 42,385 32,893 41,658 23,294 40,694 2011 50,554 39,647 47,489 43,770 41,596 34,083 40,937 23,752 37,378 2012 53,677 41,536 50,985 46,720 41,542 35,680 38,993 23,041 34,469 2013 57,067 43,665 52,210 48,365 45,357 35,774 37,636 22,804 40,982 2014 62,428 49,021 48,734 49,523 52388 39800 37411 25209 46290 2015 63,422 48,711 48,636 51,086 49572 36495 40306 25782 49134 2016 62,094 50,346 45,879 54,224 44353 41153 42909 21480 44047

Support Existing Communities 2008 473 602 899 510 834 386 663 1381 938 2009 671 1125 1526 724 845 547 1126 695 472 2010 1140 1761 2011 1256 681 950 1506 1161 788 2011 1140 2169 1863 1166 596 966 1432 935 586 2012 1031 1866 1701 1069 624 903 1482 1288 855 2013 1034 1586 1556 1010 725 883 1325 1032 723 2014 904 1398 1655 884 571 772 1409 1049 735 2015 859 1156 1521 839 525 734 1295 1024 718 2016 616 1138 1570 602 565 526 1337 1254 878